Research On Construction Method Of Deep Perception System For Health Condition Assessment Of Electrical Equipment In Substations

In 2020,Proposals for formulating the 14 th Five-Year Plan(2021-2025)for National Economic and Social Development and the Long-Range Objectives Through the Year 2035 have specifically defined that advanced technology marked by Artificial Intelligence will gradually become the foundation of country security and development.There is no doubt for the largest artificial system on earth——the power system to be the main area where Artificial Intelligence plays an important role.Thus,Artificial Intelligence starts with perception.Substations are the very important facilities in the power system and most electrical equipment in substations work in poor conditions outside year after year.Therefore,it is especially necessary to apply technologies related to deep perception to the health condition assessment of electrical equipment in substations.In recent years,State Grid Corporation of China have actively promoted the research and application of intelligent operation and maintenance for electrical equipment in substations.Accordingly,the requirements for health condition assessment will be more precise and smarter.Based on this background,the main research of this thesis focuses on the deep perception system for health condition assessment of electrical equipment in substations aiming to contribute an optional new idea.The research work and results achieved in the thesis are mainly reflected in the following four aspects:(1)By summarizing the present situation of health condition assessment and demand of various electrical equipment in substations,based on edge computing and cloud computing,the layered distributed structure is adopted to construct the deep perception system.Furthermore,Agents are set up on different sensor nodes,edge computing nodes and cloud nodes so as to realize the synergy between Multi-Agents in the deep perception system.The research work in this part on the construction of the deep perception system is the foundation for subsequent research.(2)Aiming at the spatial deployment of general sensing units in the deep perception system,based on traditional set covering problem,two mathematical models suitable for the spatial deployment of general sensing units in substations are proposed.The concept of perceiving value and perceiving resolution are introduced into the mathematical models so as to transform the three-dimensional space deployment problem into the two-dimensional grid plane.The solution and comparison of calculation examples show that the optimal fitness model proposed in this paper can effectively guide general sensing units to strengthen the perception of electrical equipment areas in the process of spatial deployment,and achieve the goal of full coverage while reducing the perception redundancy to save costs.Furthermore,a deployment idea of layered rasterization in three-dimensional space is also proposed by integrating the substation inspection robot into the deep perception system.(3)Aiming at the problem of data collection and transmission in deep perception system,a technical scheme based on compressed sensing agent is proposed and the corresponding model is built.Considering the characteristics of deep perception in substations,DWang T-4sparse basis and sparse random matrix are selected.Through analysis and simulation,the range of the signal length and the compressed measuring dimension suitable for the proposed technical scheme is obtained.Furthermore,a compressed sensing reconstruction algorithm CSRAFE based on frequency estimation is designed.The simulation proves that the reconstruction accuracy of this algorithm is significantly improved compared to the other four algorithms: OMP,Ca So MP,SL0 and NSL0.(4)Aiming at the health condition assessment of electrical equipment in substations,combining the concept of edge cloud collaboration,the perception data in the entire deep perception system is divided into two categories: the active-uploading data from the edge terminal and the inquire-response data between the edge and cloud so as to clarify the task division and collaboration mechanism in the process of health condition assessment of electrical equipment in the deep perception system.